Just got home after a weekend in San Francisco with the family, and then a customer presentation at the Google office today. Our reading list has a lot about the modern SDLC and the role of agent loops.
[article] Why we’re bullish on loops. Agents are capable of running longer, and doing more complex work. So kicking off agents that can prompt agents is a viable approach in some cases.
[article] The Case Against Building Your Own Agent Platform. Right now? You’ll likely build something incomplete or incorrect for what agents need. That’s ok, it’s a fluid market. Don’t prematurely over-platform your stack.
[article] 3 Forces Are Redefining the Transition from Manager to Leader. Advice from 2012 needed a refresh. Here’s a very good look at what it means to evolve into an organizational leader.
[blog] Scaling the Next Generation of Global Innovation: How Google Supports Top Startups Around the World. We’ve got one of the most successful startup accelerators that you’ve never heard of. Startups know it, but we’re doing some work to help others aware of it too.
[blog] AI Sped Up Coding Faster Than It Sped Up Delivery. It’s the handoffs. Always the handoffs. Think about cycle time and flow in order to identify bottlenecks faster.
[article] Autonomous Long-Running Coding Agents. Go from prompts, to goals. This makes your verifiers or evaluators even more important. I really liked this overview of the new workflow.
[blog] The New Software Lifecycle. More here on the new developer workflow. It also has a useful image showing the rapid transition from autocomplete to autonomous agents.
[article] What is GLM-5.2? Z.ai targets coding agents. Here’s a very impressive open model that’s getting some serious traction already.
[blog] Building the agentic enterprise. Terrific look at what really changes for enterprises who pursue agents, and what to expect on the journey.
[blog] From Feature Branches to Agent Branches ft. Antigravity. This developer likes how we used git worktrees in Antigravity to simplify context switching.
[article] Doubling the productivity of your engineering team using AI. That’s the dream. Very strong takeaways in this recap of a podcast interview.
[article] Backporting bug fixes is dead, Project Valkey now sends in the bots. Desperation can lead to productive implementations of AI. For example, keeping a handle on bug fixes or PRs in open source projects isn’t sustainable for most human maintainers.
[article] IT operating models remain unfit for the AI era, CEOs say. The first step is recognizing it. The second is being courageous enough to risk fixing it.
Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:
Richard Seroter's Architecture Musings 
Leave a comment